Comparison and interpretation report of the OECD international standard problem no.45 exercise (QUENCH-06)

Abstract

The International Standard Problem (ISP) No. 45 is part of the overall ISP program of the OECD/NEA and is dedicated to the behavior of heat-up and delayed reflood of fuel elements in nuclear reactors during a hypotheticalaccident. ISP-45 is related to the out-of-pile bundle quench experiment QUENCH-06, performed at Forschungszentrum Karlsruhe (FZK), Germany, on December 13, 2000. Special attention was paid to hydrogen production. To assess the ability of severe accident codes to simulate processes during core heat-up and reflood at temperatures above 2000 K, the behavior of the bundle during the whole experiment should be calculated on the basis of the necessary experimental initial and boundary conditions, but without knowing further experimental details. In this so-called blind phase 21 participants from 15 nations contributed with 8 different code systems (ATHLET-CD, ICAREICATHARE, IMPACT/SAMPSON, GENFLO, MAAP, MELCOR, SCDAPSIM, SCDAP-3D). Additionally, post-test calculations using the in-house version SCDAP/RELAP5 mod3.2.irs are used for comparison. After the end of the blind phase all measured data were made available and the participants were invited to deliver a second calculation, where this knowledge could be used (so-called open phase). In this report, results of the blind calculations are presented, analyzed, and compared to experimental data. During heat-up most results do not deviate significantly from one another, except as a consequence of some obvious user errors, so that a definition of a mainstream is justified. For the quench phase the lack of adequate hydraulic modeling becomes obvious: some participants could not match the observed cool-down rates, others had to use very fine meshes to compensate code deficiencies. To overcome this insufficiency some newly developed reflood models were used in MAAP and MELCOR. In QUENCH-06, oxide layers were thick enough to protect the cladding from melting and failure below 2200 K, so that no massive hydrogen release during reflood was found. This behavior could be simulated by most of the codes with commonly used oxidation models, if no shattering options were used arbitrarily. With respect to calculated hydrogen production the mainstream shows a spreading of ′ 15 % prior to reflood initiation and a range of ′ 40 % after reflood. However, a group of SCDAPSIM users activated an extreme shattering option, which overestimates the produced hydrogen mass by a factor of 5. In the mainstream, most of the participants calculated correctly that no bundle damage occurred, whereas others calculate slight material relocations, mainly due to overestimation of the cladding temperatures. However, detailed inspection showed that the codes still have difficulties to predict correctly the bundle initial conditions for reflood. Another surprising finding was that the energy balance has to be checked prior to further interpretation of the results. Lacking user experience and problems to model the QUENCH facility adequately was a main reason for larger deviations. In the open phase 9 participants delivered results in time, a further participant with some delay; lacking manpower or time was mentioned as main reason for not participating in the open phase. The results show that their codes are able to simulate adequately the QUENCH-06 experiment. Some participants performed successful error corrections as well as code improvements.